RESEARCH ARTICLE
Designed Amino Acid Feed in Improvement of Production and Quality Targets of a Therapeutic Monoclonal Antibody Fatemeh Torkashvand1, Behrouz Vaziri1*, Shayan Maleknia2, Amir Heydari3, Manouchehr Vossoughi3, Fatemeh Davami1, Fereidoun Mahboudi1* 1 Biotechnology Research Center, Pasteur Institute of Iran, Tehran, Iran, 2 Process Development Department, Aryogen Biopharma Inc., Alborz, Iran, 3 Department of Chemical & Petroleum Engineering, Biochemical & Bioenvironmental Research Center Sharif University of Technology, Tehran, Iran * [email protected] (BV); [email protected] (FM)
Abstract OPEN ACCESS Citation: Torkashvand F, Vaziri B, Maleknia S, Heydari A, Vossoughi M, Davami F, et al. (2015) Designed Amino Acid Feed in Improvement of Production and Quality Targets of a Therapeutic Monoclonal Antibody. PLoS ONE 10(10): e0140597. doi:10.1371/journal.pone.0140597 Editor: Matthias Johannes Schnell, Thomas Jefferson University, UNITED STATES Received: July 1, 2015 Accepted: September 27, 2015 Published: October 19, 2015 Copyright: © 2015 Torkashvand et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Cell culture feeds optimization is a critical step in process development of pharmaceutical recombinant protein production. Amino acids are the basic supplements of mammalian cell culture feeds with known effect on their growth promotion and productivity. In this study, we reported the implementation of the Plackett-Burman (PB) multifactorial design to screen the effects of amino acids on the growth promotion and productivity of a Chinese hamster ovary DG-44 (CHO-DG44) cell line producing bevacizumab. After this screening, the amino acid combinations were optimized by the response surface methodology (RSM) to determine the most effective concentration in feeds. Through this strategy, the final monoclonal antibody (mAb) titre was enhanced by 70%, compared to the control group. For this particular cell line, aspartic acid, glutamic acid, arginine and glycine had the highest positive effects on the final mAb titre. Simultaneously, the impact of the designed amino acid feed on some critical quality attributes of bevacizumab was examined in the group with highest productivity. The product was analysed for N-glycan profiles, charge variant distribution, and low molecular weight forms. The results showed that the target product quality has been improved using this feeding strategy. It was shown how this strategy could significantly diminish the time and number of experiments in identifying the most effective amino acids and related concentrations in target product enhancement. This model could be successfully applied to other components of culture media and feeds.
Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Funding: This work was funded by Aryogen Biopharma Inc. The funder provided support in the form of salaries for S. A., but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of this author are articulated in the ‘author contributions’ section. Competing Interests: Aryogen Biopharma Inc. provided support in the form of salaries for S. A. This
Introduction Recent progress in cell culture technology for recombinant CHO cells has led to substantial enhancements in target protein’s production [1, 2]. This advancement in the production yields is mostly due to the extension of stable high producers through vector design and host cell engineering methods, as well as medium optimization and process development [3–5].
PLOS ONE | DOI:10.1371/journal.pone.0140597 October 19, 2015
1 / 21
Designed Amino Acid Feed in mAb Titre and Quality
does not alter the authors' adherence to PLOS ONE policies on sharing data and materials.
The formulation of media and feeds is an important phase of the process development. Manufacturers spend extensive efforts and time on the optimization of culture media and feeds as a basic development process for each cell line [4, 6]. However, no generic procedure exists for cell culture medium and feed optimization, and reports on the optimization of culture media or feeding strategies for CHO cell cultures are limited in number. A valuable start to optimizing the medium or feed is to focus on basic groups of ingredients composing mammalian cell culture media. Traditional optimization methods, such as the titration of single components, are reliable, but can be labour intensive and time consuming [7–9]. Design Of Experiment (DOE) methods are useful in some optimization steps for decreasing the number of experiments [4]. In fact, any strategy for decreasing the large number of components in media or feeds to those that exert major effects leads to substantial savings in time and cost [10, 11]. Screening experimental designs with these components at low and high ranges of concentration is helpful for determining which components have important effects on productivity. After determining the critical components, optimization of their amount is necessary. This step can be performed with DOE methods, such as response surface methodology (RSM). Amino acids are amongst the most basic nutrients for promoting cell growth and progress in productivity. They are nitrogen sources and the building blocks of proteins as well as mediators of numerous metabolic pathways [3, 12–14]. Although amino acid supplementation is recognized as one of the vital components in cell culture medium design and optimization, only a few reports have been focused on it [3, 8, 12–17]. In this study, we determined the critical amino acids for enhancement the target mAb production in a specific CHO cell line using Plackett-Burman design. After finding the critical amino acids, the concentration of these components in feed was optimized by RSM using a Box-Bencken design. The best group from RSM analysis was selected to explore the effect of designed amino acid feed on the main quality attributes of bevacizumab. N-glycan profiles, charge variant distribution, and low molecular weight forms are the main quality attributes for a monoclonal antibody and significant changes in these properties impact the mAb efficacy [18, 19]. The mentioned attributes are dependent in host cell line, clone, process conditions, and media composition. Therefore, analysing these product quality characteristics in process development is very useful to match the desired quality target product profile [20]. To our knowledge, this is the first report using the Plackett-Burman experimental design and RSM to investigate the most effective amino acids and optimize them in feed in order to enhance the recombinant mAb production. These strategy could be used for the optimization of other feed components (i.e., sugar, lipids, vitamins, and trace elements) to produce more developed feeds.
Materials and Methods Cell line and culture media Recombinant CHO cell producing bevasizumab, a mAb against vascular endothelial growth factor A, was kindly provided by Aryogen Biopharma (Alborz, Iran). The original source of CHO DG44 cell line was purchased from gibco (Catalog no: A10971-01). The basal culture medium was CDM4CHO (Hyclone laboratories, Utah, USA) which was supplemented with 6mM L-glutamine (Lonza, Verviers, Belgium) at the moment of preparation in accordance to the recommendations from the supplier.
Culture conditions CHO cells were cultured in a 500ml shaker flask with an effective volume of 100 ml, incubated at 37°C with a 5% solution of CO2, and agitated at 80 rpm. In the middle of logarithmic phase,
PLOS ONE | DOI:10.1371/journal.pone.0140597 October 19, 2015
2 / 21
Designed Amino Acid Feed in mAb Titre and Quality
the temperature shifted to 32°C. Each shaker flask was inoculated with an approximate cell density of 5×105 cell/ml. The amino acid feeds were added as multiple discrete additions to cultures on day 3, 5 and 7. All amino acid powders were purchased from HiMedia Laboratories Company (Mumbai, India).
Cell quantification Cell density and cell viability were determined by the Trypan blue exclusion method using a Neubauer cytometer. The harvested cells were centrifuged at 4000 rpm for 5 min and the supernatants were stored at -70°C for later analysis.
Amino acid quantification For quantification of amino acids, they were derivatized by OPA (Pickering Laboratories, CA, USA), separated by HPLC and detected by a florescence detector [21]. Consumption rates for each amino acid were determined from the analysis of the concentration profiles at the time observed in the 500 ml shaker flask. For that purpose, the consumption rates were assumed to follow the simple model d[AA]/dt = -qAA, where [AA] is a particular amino acid concentration, t is time, d[AA]/dt denotes the derivative of a particular amino acid concentration with respect to time, and qAA is the consumption rate of that amino acid. Therefore, qAA for each amino acid was calculated from the slope of the linear portion of the corresponding plot of concentration versus time.
Monoclonal antibody quantification The final concentration of the monoclonal antibody in the samples was determined by the MAbPac protein A affinity column (Thermo scientific, CA, USA). A simple two-buffer system was used for column operation. The equilibration buffer or Buffer A (5% PBS, 0.15M sodium chloride (Merck, Darmstadt, Germany) and 5% acetonitryl (Merck, Darmstadt, Germany)) was adjusted to pH 7.5 with ortophosphoric acid. The elution buffer (Buffer B) was the same buffer, but was adjusted to pH 2.5 with orthophosphoric acid. The elution gradient was programmed from 0% to 100% of elution buffer B in approximately 5 min. The standard curve generated with the purified monoclonal IgG and the mAb quantification was performed based on this curve.
Experimental design Screening the significant amino acids. A 20-run Plackett–Burman design was performed to explore the effect of 19 amino acids. According to this design matrix, a total of 20 trials were performed at various combinations of ‘high’ (1) and ‘low’ (-1) values of the different amino acids (Table 1), the low values were the amount of amino acids in basal medium and high values were determined with some pre-experiments. The statistical design and analysis were performed using the Design-Expert1 software version7 (Stat-EaseInc.Minneapolis, Minnesota, USA). The confidence level for significance was 95% (p F of less than 0.05 indicated that the model terms were significant. Locating optimum concentration of significant amino acids. After the selection of four significant amino acids by Plackett-Burman design, Box–Behnken statistical experimental design (a common method of RSM) was employed for the further optimization of these amino acids based on 29 sets of experiments. The levels corresponding to the center and the walls of this design were designated 0, –1 and +1. Table 2 lists various assemblies that were used in the design. The low level used in the design (designated –1) was the same as the higher level used in the Plackett-Burman design. The middle and high levels (designated 0 and +1) chosen in the Box-Bencken design were respectively 25% and 50% higher than the low levels. The final mAb titre was used as the response parameter in this step.
Glycoprofiling of IgG Ig Gglycoprofiling was carried out using N-Glycanase1 Kit (Prozyme, CA, USA) for quantifying the total of each N-glycans. Digestion and labelling of N-glycans from IgG was performed according to the manufacturer’s instructions. Labelled glycans were buffered in 70% v/v aqueous acetonitrile prior to the HPLC analysis using HPLC system equipped with fluorescence detector (KNAUER) and TSK-gel Amide-80, 4.6× 230 mm column (TOSOH biosience, Stuttgart, Germany). The mobile phases used were solvent A (100% acetonitrile) and solvent B (0.25M ammonium formate (Sigma- Aldrich, MO, USA) with pH = 4.4). Elution of the sample was performed using the following flow rate and gradient: T0 min = 30% B with flow rate = 0.4mL/min, T100 min = 40% B with flow rate = 0.4 mL/min, T103 min = 100% B with flowrate 1 mL/min, T113 min 100% B with flow rate 1 mL/min,T114 min 30% B with flow rate 1mL/min, T119 min 30% B with flow rate 1 mL/min, T120 min 30% B with flowrate 0.4 mL/min. Fluorescence
PLOS ONE | DOI:10.1371/journal.pone.0140597 October 19, 2015
4 / 21
Designed Amino Acid Feed in mAb Titre and Quality
Table 2. Box-Bencken design matrix, the low level used in the design (–1) was the same as the higher level used in the Plackett-Burman design. The middle and high levels (0 and +1) were respectively 25% and 50% higher than the low levels. Run
Asp
Glu
Arg
Gly
1
0
0
1
-1
2
1
0
0
-1
3
0
-1
1
0
4
0
-1
-1
0
5
0
1
-1
0
6
-1
-1
0
0
7
0
-1
0
-1
8
-1
0
0
-1
9
0
1
0
1
10
1
0
1
0
11
-1
1
0
0
12
1
1
0
0
13
0
0
0
0
14
-1
0
0
1
15
0
0
0
0
16
0
0
0
0
17
-1
0
1
0
18
-1
0
-1
0
19
1
0
0
1
20
0
-1
0
1
21
0
0
1
1
22
0
0
0
0
23
0
1
0
-1
24
0
0
0
0
25
1
0
-1
0
26
0
0
-1
1
27
0
1
1
0
28
1
-1
0
0
29
0
0
-1
-1
doi:10.1371/journal.pone.0140597.t002
detection was performed at Ex = 360 nm and Em = 425 nm. Analysis of the chromatogram was performed with ChromGate software (KNAUER). Relative quantification was performed using peak area and peak assignment based on retention time of known standards.
Charge heterogeneity profiling of IgG The IEX chromatography was performed on a liquid chromatograph (KNAUER,) Detection was performed at 280 nm. Flow rate was 0.8 mL/min, the injection volume was 80μL and the column compartment temperature was set at 25°C. Instrument control and data acquisition were performed using ChromGate software. The monolithic IEX column used was ProPac WCX-10, 4 mm × 250 mm (Thermo Scientific, USA). The mobile phases used were mobile phase A (0.01 M Sodium phosphate (Merck, Darmstadt, Germany) buffer pH 6.6) and mobile phase B (0.01 M Sodium phosphate buffer + 0.1M Sodium chloride pH 6.6). The elution was performed by an ascending gradient from 30% to 100% eluent B followed by isocratic elution for 2 min before returning the eluent composition to the starting condition (100% eluent A).
PLOS ONE | DOI:10.1371/journal.pone.0140597 October 19, 2015
5 / 21
Designed Amino Acid Feed in mAb Titre and Quality
Capillary electrophoresis SDS analysis An Agilent 7100 CE instrument equipped with a photodiode array (PDA) detector (Agilent) was used to perform CE–SDS analyses. For the separation, fused-silica capillaries of 50 μm inner diameter with a total length of 30.2 cm and an effective length of 20 cm was used. The capillary was conditioned at 4 bar with 0.1 M NaOH, 0.1 M HCl, and deionized water for 3, 1, and 1 min, respectively, followed by filling the capillary with the running gel buffer for 10 min at 4bar. Sample loading was performed electrokinetically at 5 kV (reverse polarity) for 20 s. The sample analysis was performed by applying 15 kV (reverse polarity) for 30 min, and the detection wavelength was 220 nm. The capillary and autosampler were maintained at 25°C and 10°C respectively during the separation. Samples were prepared by mixing 100 μg of protein with either 5 μl of 0.25M idoacetamide (GE Healthcare, Backing hamshire, Uk) for non-reduced samples or 5 μl of β-mercaptoethanol (Sigma Aldrich, MO, USA) for reduced samples and the total volume of the sample was then adjusted to 100 μl by adding an appropriate amount of SDS sample buffer (0.1M Tris–HCl (pH 9.0) and 1% SDS) in a microcentrifuge tube. The mixture was vortexed and centrifuged briefly to bring the contents down to the bottom of the tube. Samples were heated for 10 min at 70°C before analysis.
Results and Discussion Amino acid analysis The consumption rate for each amino acid can be calculated as d[AA]/dt = -qAA. The qAA for each amino acid can be calculated from the slope of the linear section of the matching concentration versus time graph [22]. If amino acids are added at inappropriate times and concentrations, undesirable by-products accumulates and osmolarity increases [23]. The amino acid analysis was performed to determine the appropriate optimal time to add amino acid feeds to the basal medium. Based on this analysis, it was found that day 3 was a good starting point. Furthermore, through some pre-experiments the maximum concentration of amino acids for the Plackett-Burman design was determined (data not shown). However, as reported, the particular effect of each amino acid changes across the different CHO cell lines and different products [3, 24].
Plackett–Burman design The Plackett–Burman design and its analysis were performed with the design-expert1 software. As presented in Table 3, the effect parameter is the change in the final mAb titre response as the factor (amino acid concentration) changes from its low (-1) level to its high (+1) level. The sum of square for a term is the sum of squared deviations from the mean due to the effect of the term. The contribution% is calculated by the total sum of squares (SS) and then dividing each individual SS by the total SS, multiplied by 100. When all terms have the same degrees of freedom, the contribution% can be used to determine which terms are larger contributors than others. According to Table 3, based on the contribution% and effect parameters, Asp, Glu, Arg and Gly have the highest positive effect on the final mAb titre response, and Asn, Ser and Phe have the highest negative effect. Table 4 represents the effects of amino acids on the IVCC response. The values of the effect, sum square and contribution% parameters are the same as those of parameters in Table 3. Based on the contribution% and effect parameters, Asp, Glu and Gly have the highest positive effect on the integral viable cell concentration response, and Asn, Val and Phe have the highest negative effect.
PLOS ONE | DOI:10.1371/journal.pone.0140597 October 19, 2015
6 / 21
Designed Amino Acid Feed in mAb Titre and Quality
Table 3. The effects of amino acids on the final mAb titre, the effect parameter is the change in the final mAb titre as the amino acid concentration changes from its low level to its high level. Sum of square is the sum of squared deviations from the mean due to the effect of the term. Contribution% determines the individual amino acid contribution to final mAb titre in comparison to others. Term
Effect
SumSquare
Asp*
8.1
328.05
24.99
Glu*
3.9
76.05
5.79
Asn*
-9.9
490.05
37.34
Ser*
-5.10
130.05
9.91
His*
0.90
4.05
0.31
Arg*
2.90
42.05
3.2
2.70
Gly*
Contribution%
36.45
2.78
Thr
-0.3
0.45
0.034
Ala
-0.1
0.050
Tyr*
3.809E-003
1.90
18.05
1.38
Met*
2.10
22.05
3.68
Val*
-3.50
61.25
4.67
Phe*
-3.70
68.45
5.22
Iso*
-1.10
6.05
0.46
Leu
0.3
0.45
0.034
Lys*
-1.10
6.05
0.46
Cys*
1.30
8.45
0.64
Trp
0.1
0.05
3.809E-003
-1.70
14.45
Pro*
1.10
* Amino acids in bold with a contribution% greater than 0.1 were chosen for the model in ANOVA. doi:10.1371/journal.pone.0140597.t003
Table 5 summarizes the ANOVA results for the amino acids which were chosen based on the contribution% for the final mAb titre. Amino acids with % of contribution higher than 0.1 were chosen for ANOVA. P value Prob> F parameter should be less than 0.05 to be strongly significant. The significance of the proposed model for the final mAb titre response was indicated by the F-value (349.75) and a low probability value (P-value < 0.0001). The proposed model for the final mAb titre response is expressed as an empirical first order polynomial equation in terms of fifteen variables in Eq 1: final mAb titre ¼ þ 52:85 þ 4:05A þ 1
.95
B 4:95C 2:55D þ 0:45E þ 1:45F þ 1:35G þ 0:95K
þ1:05L 1:75M 1:85N 0:55O 0:55Q þ 0:65R 0:85T
ð1Þ
A, B, C, D, E, F, G, K, L, M, N, O, Q, R and T are coded values of Asp, Glu, Asn, Ser, His, Arg, Gly, Tyr, Met, Val, Phe, Iso, Lys, Cys and Pro, respectively. Based on the results of Plackett-Burman analysis, the amino acids with the most pronounced effect on the final mAb titre were, in descending order by significance of effect, Asp, Asn, Ser, Glu, Phe, Val, Arg, Gly, Met, Tyr, Pro, Cys, Lys, Iso and His. For the range of concentrations which were tested in this experiment, Asn, Ser, Phe, Val, Iso and Lys were found to be harmful for the final mAb titre. However, the effects of Asp, Glu, Arg, Gly, Met, Tyr, Cys and His were positive. The other amino acids did not have a significant effect on the final mAb titre.
PLOS ONE | DOI:10.1371/journal.pone.0140597 October 19, 2015
7 / 21
Designed Amino Acid Feed in mAb Titre and Quality
Table 4. The effects of amino acids on Integral Viable Cell Concentration, the effect parameter is the change in IVCC as the amino acid concentration changes from its low level to its high level. SumSquare is the sum of squared deviations from the mean of the effect parameter for each amino acid. Contribution% determines the individual amino acid contribution to IVCC response. Term
Effect
SumSquare
Contribution%
Asp*
637.20
2.030E+003
20.66
Glu*
433.00
9.374E+005
9.54
Asn*
-794.40
3.155E+006
32.11
Ser
53.60
14364.80
0.15
His
37
6845.00
0.070
Arg
87.80
38544.20
0.39
Gly*
357.20
6.380E+005
6.49
Thr*
-131.60
86592.80
0.88
Ala*
197
1.940E+005
1.97
Tyr*
159.80
1.277E+005
1.30
Met*
299.0
4.470E+005
4.55
Val*
-561.60
1.691E+006
17.21
Phe*
-176.60
1.559E+005
1.59
Iso
-47.20
1113920
0.11
Leu
35.60
6336.80
0.064
Lys
64.60
20865.80
0.21
Cys*
217.60
2.367E+005
2.41
Trp
-76.40
29184.80
0.03
Pro
8.00
320.00
3.256E-003
* Amino acids in bold with a % of contribution greater than 0.4 were chosen for the model in ANOVA. doi:10.1371/journal.pone.0140597.t004
Table 5. ANOVA results for Plackett-Burman analysis for the final mAb titre response Source Final mAb titre
Sum of square
df
Mean Square
F Value
P Value Prob> F
1311.55
15
87.44
349.75
Designed Amino Acid Feed in Improvement of Production and Quality Targets of a Therapeutic Monoclonal Antibody Fatemeh Torkashvand1, Behrouz Vaziri1*, Shayan Maleknia2, Amir Heydari3, Manouchehr Vossoughi3, Fatemeh Davami1, Fereidoun Mahboudi1* 1 Biotechnology Research Center, Pasteur Institute of Iran, Tehran, Iran, 2 Process Development Department, Aryogen Biopharma Inc., Alborz, Iran, 3 Department of Chemical & Petroleum Engineering, Biochemical & Bioenvironmental Research Center Sharif University of Technology, Tehran, Iran * [email protected] (BV); [email protected] (FM)
Abstract OPEN ACCESS Citation: Torkashvand F, Vaziri B, Maleknia S, Heydari A, Vossoughi M, Davami F, et al. (2015) Designed Amino Acid Feed in Improvement of Production and Quality Targets of a Therapeutic Monoclonal Antibody. PLoS ONE 10(10): e0140597. doi:10.1371/journal.pone.0140597 Editor: Matthias Johannes Schnell, Thomas Jefferson University, UNITED STATES Received: July 1, 2015 Accepted: September 27, 2015 Published: October 19, 2015 Copyright: © 2015 Torkashvand et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Cell culture feeds optimization is a critical step in process development of pharmaceutical recombinant protein production. Amino acids are the basic supplements of mammalian cell culture feeds with known effect on their growth promotion and productivity. In this study, we reported the implementation of the Plackett-Burman (PB) multifactorial design to screen the effects of amino acids on the growth promotion and productivity of a Chinese hamster ovary DG-44 (CHO-DG44) cell line producing bevacizumab. After this screening, the amino acid combinations were optimized by the response surface methodology (RSM) to determine the most effective concentration in feeds. Through this strategy, the final monoclonal antibody (mAb) titre was enhanced by 70%, compared to the control group. For this particular cell line, aspartic acid, glutamic acid, arginine and glycine had the highest positive effects on the final mAb titre. Simultaneously, the impact of the designed amino acid feed on some critical quality attributes of bevacizumab was examined in the group with highest productivity. The product was analysed for N-glycan profiles, charge variant distribution, and low molecular weight forms. The results showed that the target product quality has been improved using this feeding strategy. It was shown how this strategy could significantly diminish the time and number of experiments in identifying the most effective amino acids and related concentrations in target product enhancement. This model could be successfully applied to other components of culture media and feeds.
Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Funding: This work was funded by Aryogen Biopharma Inc. The funder provided support in the form of salaries for S. A., but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of this author are articulated in the ‘author contributions’ section. Competing Interests: Aryogen Biopharma Inc. provided support in the form of salaries for S. A. This
Introduction Recent progress in cell culture technology for recombinant CHO cells has led to substantial enhancements in target protein’s production [1, 2]. This advancement in the production yields is mostly due to the extension of stable high producers through vector design and host cell engineering methods, as well as medium optimization and process development [3–5].
PLOS ONE | DOI:10.1371/journal.pone.0140597 October 19, 2015
1 / 21
Designed Amino Acid Feed in mAb Titre and Quality
does not alter the authors' adherence to PLOS ONE policies on sharing data and materials.
The formulation of media and feeds is an important phase of the process development. Manufacturers spend extensive efforts and time on the optimization of culture media and feeds as a basic development process for each cell line [4, 6]. However, no generic procedure exists for cell culture medium and feed optimization, and reports on the optimization of culture media or feeding strategies for CHO cell cultures are limited in number. A valuable start to optimizing the medium or feed is to focus on basic groups of ingredients composing mammalian cell culture media. Traditional optimization methods, such as the titration of single components, are reliable, but can be labour intensive and time consuming [7–9]. Design Of Experiment (DOE) methods are useful in some optimization steps for decreasing the number of experiments [4]. In fact, any strategy for decreasing the large number of components in media or feeds to those that exert major effects leads to substantial savings in time and cost [10, 11]. Screening experimental designs with these components at low and high ranges of concentration is helpful for determining which components have important effects on productivity. After determining the critical components, optimization of their amount is necessary. This step can be performed with DOE methods, such as response surface methodology (RSM). Amino acids are amongst the most basic nutrients for promoting cell growth and progress in productivity. They are nitrogen sources and the building blocks of proteins as well as mediators of numerous metabolic pathways [3, 12–14]. Although amino acid supplementation is recognized as one of the vital components in cell culture medium design and optimization, only a few reports have been focused on it [3, 8, 12–17]. In this study, we determined the critical amino acids for enhancement the target mAb production in a specific CHO cell line using Plackett-Burman design. After finding the critical amino acids, the concentration of these components in feed was optimized by RSM using a Box-Bencken design. The best group from RSM analysis was selected to explore the effect of designed amino acid feed on the main quality attributes of bevacizumab. N-glycan profiles, charge variant distribution, and low molecular weight forms are the main quality attributes for a monoclonal antibody and significant changes in these properties impact the mAb efficacy [18, 19]. The mentioned attributes are dependent in host cell line, clone, process conditions, and media composition. Therefore, analysing these product quality characteristics in process development is very useful to match the desired quality target product profile [20]. To our knowledge, this is the first report using the Plackett-Burman experimental design and RSM to investigate the most effective amino acids and optimize them in feed in order to enhance the recombinant mAb production. These strategy could be used for the optimization of other feed components (i.e., sugar, lipids, vitamins, and trace elements) to produce more developed feeds.
Materials and Methods Cell line and culture media Recombinant CHO cell producing bevasizumab, a mAb against vascular endothelial growth factor A, was kindly provided by Aryogen Biopharma (Alborz, Iran). The original source of CHO DG44 cell line was purchased from gibco (Catalog no: A10971-01). The basal culture medium was CDM4CHO (Hyclone laboratories, Utah, USA) which was supplemented with 6mM L-glutamine (Lonza, Verviers, Belgium) at the moment of preparation in accordance to the recommendations from the supplier.
Culture conditions CHO cells were cultured in a 500ml shaker flask with an effective volume of 100 ml, incubated at 37°C with a 5% solution of CO2, and agitated at 80 rpm. In the middle of logarithmic phase,
PLOS ONE | DOI:10.1371/journal.pone.0140597 October 19, 2015
2 / 21
Designed Amino Acid Feed in mAb Titre and Quality
the temperature shifted to 32°C. Each shaker flask was inoculated with an approximate cell density of 5×105 cell/ml. The amino acid feeds were added as multiple discrete additions to cultures on day 3, 5 and 7. All amino acid powders were purchased from HiMedia Laboratories Company (Mumbai, India).
Cell quantification Cell density and cell viability were determined by the Trypan blue exclusion method using a Neubauer cytometer. The harvested cells were centrifuged at 4000 rpm for 5 min and the supernatants were stored at -70°C for later analysis.
Amino acid quantification For quantification of amino acids, they were derivatized by OPA (Pickering Laboratories, CA, USA), separated by HPLC and detected by a florescence detector [21]. Consumption rates for each amino acid were determined from the analysis of the concentration profiles at the time observed in the 500 ml shaker flask. For that purpose, the consumption rates were assumed to follow the simple model d[AA]/dt = -qAA, where [AA] is a particular amino acid concentration, t is time, d[AA]/dt denotes the derivative of a particular amino acid concentration with respect to time, and qAA is the consumption rate of that amino acid. Therefore, qAA for each amino acid was calculated from the slope of the linear portion of the corresponding plot of concentration versus time.
Monoclonal antibody quantification The final concentration of the monoclonal antibody in the samples was determined by the MAbPac protein A affinity column (Thermo scientific, CA, USA). A simple two-buffer system was used for column operation. The equilibration buffer or Buffer A (5% PBS, 0.15M sodium chloride (Merck, Darmstadt, Germany) and 5% acetonitryl (Merck, Darmstadt, Germany)) was adjusted to pH 7.5 with ortophosphoric acid. The elution buffer (Buffer B) was the same buffer, but was adjusted to pH 2.5 with orthophosphoric acid. The elution gradient was programmed from 0% to 100% of elution buffer B in approximately 5 min. The standard curve generated with the purified monoclonal IgG and the mAb quantification was performed based on this curve.
Experimental design Screening the significant amino acids. A 20-run Plackett–Burman design was performed to explore the effect of 19 amino acids. According to this design matrix, a total of 20 trials were performed at various combinations of ‘high’ (1) and ‘low’ (-1) values of the different amino acids (Table 1), the low values were the amount of amino acids in basal medium and high values were determined with some pre-experiments. The statistical design and analysis were performed using the Design-Expert1 software version7 (Stat-EaseInc.Minneapolis, Minnesota, USA). The confidence level for significance was 95% (p F of less than 0.05 indicated that the model terms were significant. Locating optimum concentration of significant amino acids. After the selection of four significant amino acids by Plackett-Burman design, Box–Behnken statistical experimental design (a common method of RSM) was employed for the further optimization of these amino acids based on 29 sets of experiments. The levels corresponding to the center and the walls of this design were designated 0, –1 and +1. Table 2 lists various assemblies that were used in the design. The low level used in the design (designated –1) was the same as the higher level used in the Plackett-Burman design. The middle and high levels (designated 0 and +1) chosen in the Box-Bencken design were respectively 25% and 50% higher than the low levels. The final mAb titre was used as the response parameter in this step.
Glycoprofiling of IgG Ig Gglycoprofiling was carried out using N-Glycanase1 Kit (Prozyme, CA, USA) for quantifying the total of each N-glycans. Digestion and labelling of N-glycans from IgG was performed according to the manufacturer’s instructions. Labelled glycans were buffered in 70% v/v aqueous acetonitrile prior to the HPLC analysis using HPLC system equipped with fluorescence detector (KNAUER) and TSK-gel Amide-80, 4.6× 230 mm column (TOSOH biosience, Stuttgart, Germany). The mobile phases used were solvent A (100% acetonitrile) and solvent B (0.25M ammonium formate (Sigma- Aldrich, MO, USA) with pH = 4.4). Elution of the sample was performed using the following flow rate and gradient: T0 min = 30% B with flow rate = 0.4mL/min, T100 min = 40% B with flow rate = 0.4 mL/min, T103 min = 100% B with flowrate 1 mL/min, T113 min 100% B with flow rate 1 mL/min,T114 min 30% B with flow rate 1mL/min, T119 min 30% B with flow rate 1 mL/min, T120 min 30% B with flowrate 0.4 mL/min. Fluorescence
PLOS ONE | DOI:10.1371/journal.pone.0140597 October 19, 2015
4 / 21
Designed Amino Acid Feed in mAb Titre and Quality
Table 2. Box-Bencken design matrix, the low level used in the design (–1) was the same as the higher level used in the Plackett-Burman design. The middle and high levels (0 and +1) were respectively 25% and 50% higher than the low levels. Run
Asp
Glu
Arg
Gly
1
0
0
1
-1
2
1
0
0
-1
3
0
-1
1
0
4
0
-1
-1
0
5
0
1
-1
0
6
-1
-1
0
0
7
0
-1
0
-1
8
-1
0
0
-1
9
0
1
0
1
10
1
0
1
0
11
-1
1
0
0
12
1
1
0
0
13
0
0
0
0
14
-1
0
0
1
15
0
0
0
0
16
0
0
0
0
17
-1
0
1
0
18
-1
0
-1
0
19
1
0
0
1
20
0
-1
0
1
21
0
0
1
1
22
0
0
0
0
23
0
1
0
-1
24
0
0
0
0
25
1
0
-1
0
26
0
0
-1
1
27
0
1
1
0
28
1
-1
0
0
29
0
0
-1
-1
doi:10.1371/journal.pone.0140597.t002
detection was performed at Ex = 360 nm and Em = 425 nm. Analysis of the chromatogram was performed with ChromGate software (KNAUER). Relative quantification was performed using peak area and peak assignment based on retention time of known standards.
Charge heterogeneity profiling of IgG The IEX chromatography was performed on a liquid chromatograph (KNAUER,) Detection was performed at 280 nm. Flow rate was 0.8 mL/min, the injection volume was 80μL and the column compartment temperature was set at 25°C. Instrument control and data acquisition were performed using ChromGate software. The monolithic IEX column used was ProPac WCX-10, 4 mm × 250 mm (Thermo Scientific, USA). The mobile phases used were mobile phase A (0.01 M Sodium phosphate (Merck, Darmstadt, Germany) buffer pH 6.6) and mobile phase B (0.01 M Sodium phosphate buffer + 0.1M Sodium chloride pH 6.6). The elution was performed by an ascending gradient from 30% to 100% eluent B followed by isocratic elution for 2 min before returning the eluent composition to the starting condition (100% eluent A).
PLOS ONE | DOI:10.1371/journal.pone.0140597 October 19, 2015
5 / 21
Designed Amino Acid Feed in mAb Titre and Quality
Capillary electrophoresis SDS analysis An Agilent 7100 CE instrument equipped with a photodiode array (PDA) detector (Agilent) was used to perform CE–SDS analyses. For the separation, fused-silica capillaries of 50 μm inner diameter with a total length of 30.2 cm and an effective length of 20 cm was used. The capillary was conditioned at 4 bar with 0.1 M NaOH, 0.1 M HCl, and deionized water for 3, 1, and 1 min, respectively, followed by filling the capillary with the running gel buffer for 10 min at 4bar. Sample loading was performed electrokinetically at 5 kV (reverse polarity) for 20 s. The sample analysis was performed by applying 15 kV (reverse polarity) for 30 min, and the detection wavelength was 220 nm. The capillary and autosampler were maintained at 25°C and 10°C respectively during the separation. Samples were prepared by mixing 100 μg of protein with either 5 μl of 0.25M idoacetamide (GE Healthcare, Backing hamshire, Uk) for non-reduced samples or 5 μl of β-mercaptoethanol (Sigma Aldrich, MO, USA) for reduced samples and the total volume of the sample was then adjusted to 100 μl by adding an appropriate amount of SDS sample buffer (0.1M Tris–HCl (pH 9.0) and 1% SDS) in a microcentrifuge tube. The mixture was vortexed and centrifuged briefly to bring the contents down to the bottom of the tube. Samples were heated for 10 min at 70°C before analysis.
Results and Discussion Amino acid analysis The consumption rate for each amino acid can be calculated as d[AA]/dt = -qAA. The qAA for each amino acid can be calculated from the slope of the linear section of the matching concentration versus time graph [22]. If amino acids are added at inappropriate times and concentrations, undesirable by-products accumulates and osmolarity increases [23]. The amino acid analysis was performed to determine the appropriate optimal time to add amino acid feeds to the basal medium. Based on this analysis, it was found that day 3 was a good starting point. Furthermore, through some pre-experiments the maximum concentration of amino acids for the Plackett-Burman design was determined (data not shown). However, as reported, the particular effect of each amino acid changes across the different CHO cell lines and different products [3, 24].
Plackett–Burman design The Plackett–Burman design and its analysis were performed with the design-expert1 software. As presented in Table 3, the effect parameter is the change in the final mAb titre response as the factor (amino acid concentration) changes from its low (-1) level to its high (+1) level. The sum of square for a term is the sum of squared deviations from the mean due to the effect of the term. The contribution% is calculated by the total sum of squares (SS) and then dividing each individual SS by the total SS, multiplied by 100. When all terms have the same degrees of freedom, the contribution% can be used to determine which terms are larger contributors than others. According to Table 3, based on the contribution% and effect parameters, Asp, Glu, Arg and Gly have the highest positive effect on the final mAb titre response, and Asn, Ser and Phe have the highest negative effect. Table 4 represents the effects of amino acids on the IVCC response. The values of the effect, sum square and contribution% parameters are the same as those of parameters in Table 3. Based on the contribution% and effect parameters, Asp, Glu and Gly have the highest positive effect on the integral viable cell concentration response, and Asn, Val and Phe have the highest negative effect.
PLOS ONE | DOI:10.1371/journal.pone.0140597 October 19, 2015
6 / 21
Designed Amino Acid Feed in mAb Titre and Quality
Table 3. The effects of amino acids on the final mAb titre, the effect parameter is the change in the final mAb titre as the amino acid concentration changes from its low level to its high level. Sum of square is the sum of squared deviations from the mean due to the effect of the term. Contribution% determines the individual amino acid contribution to final mAb titre in comparison to others. Term
Effect
SumSquare
Asp*
8.1
328.05
24.99
Glu*
3.9
76.05
5.79
Asn*
-9.9
490.05
37.34
Ser*
-5.10
130.05
9.91
His*
0.90
4.05
0.31
Arg*
2.90
42.05
3.2
2.70
Gly*
Contribution%
36.45
2.78
Thr
-0.3
0.45
0.034
Ala
-0.1
0.050
Tyr*
3.809E-003
1.90
18.05
1.38
Met*
2.10
22.05
3.68
Val*
-3.50
61.25
4.67
Phe*
-3.70
68.45
5.22
Iso*
-1.10
6.05
0.46
Leu
0.3
0.45
0.034
Lys*
-1.10
6.05
0.46
Cys*
1.30
8.45
0.64
Trp
0.1
0.05
3.809E-003
-1.70
14.45
Pro*
1.10
* Amino acids in bold with a contribution% greater than 0.1 were chosen for the model in ANOVA. doi:10.1371/journal.pone.0140597.t003
Table 5 summarizes the ANOVA results for the amino acids which were chosen based on the contribution% for the final mAb titre. Amino acids with % of contribution higher than 0.1 were chosen for ANOVA. P value Prob> F parameter should be less than 0.05 to be strongly significant. The significance of the proposed model for the final mAb titre response was indicated by the F-value (349.75) and a low probability value (P-value < 0.0001). The proposed model for the final mAb titre response is expressed as an empirical first order polynomial equation in terms of fifteen variables in Eq 1: final mAb titre ¼ þ 52:85 þ 4:05A þ 1
.95
B 4:95C 2:55D þ 0:45E þ 1:45F þ 1:35G þ 0:95K
þ1:05L 1:75M 1:85N 0:55O 0:55Q þ 0:65R 0:85T
ð1Þ
A, B, C, D, E, F, G, K, L, M, N, O, Q, R and T are coded values of Asp, Glu, Asn, Ser, His, Arg, Gly, Tyr, Met, Val, Phe, Iso, Lys, Cys and Pro, respectively. Based on the results of Plackett-Burman analysis, the amino acids with the most pronounced effect on the final mAb titre were, in descending order by significance of effect, Asp, Asn, Ser, Glu, Phe, Val, Arg, Gly, Met, Tyr, Pro, Cys, Lys, Iso and His. For the range of concentrations which were tested in this experiment, Asn, Ser, Phe, Val, Iso and Lys were found to be harmful for the final mAb titre. However, the effects of Asp, Glu, Arg, Gly, Met, Tyr, Cys and His were positive. The other amino acids did not have a significant effect on the final mAb titre.
PLOS ONE | DOI:10.1371/journal.pone.0140597 October 19, 2015
7 / 21
Designed Amino Acid Feed in mAb Titre and Quality
Table 4. The effects of amino acids on Integral Viable Cell Concentration, the effect parameter is the change in IVCC as the amino acid concentration changes from its low level to its high level. SumSquare is the sum of squared deviations from the mean of the effect parameter for each amino acid. Contribution% determines the individual amino acid contribution to IVCC response. Term
Effect
SumSquare
Contribution%
Asp*
637.20
2.030E+003
20.66
Glu*
433.00
9.374E+005
9.54
Asn*
-794.40
3.155E+006
32.11
Ser
53.60
14364.80
0.15
His
37
6845.00
0.070
Arg
87.80
38544.20
0.39
Gly*
357.20
6.380E+005
6.49
Thr*
-131.60
86592.80
0.88
Ala*
197
1.940E+005
1.97
Tyr*
159.80
1.277E+005
1.30
Met*
299.0
4.470E+005
4.55
Val*
-561.60
1.691E+006
17.21
Phe*
-176.60
1.559E+005
1.59
Iso
-47.20
1113920
0.11
Leu
35.60
6336.80
0.064
Lys
64.60
20865.80
0.21
Cys*
217.60
2.367E+005
2.41
Trp
-76.40
29184.80
0.03
Pro
8.00
320.00
3.256E-003
* Amino acids in bold with a % of contribution greater than 0.4 were chosen for the model in ANOVA. doi:10.1371/journal.pone.0140597.t004
Table 5. ANOVA results for Plackett-Burman analysis for the final mAb titre response Source Final mAb titre
Sum of square
df
Mean Square
F Value
P Value Prob> F
1311.55
15
87.44
349.75
